In this paper, we propose a family of unicyclic graphs to study robustness of network coherence quantified by the Laplacian spectrum, which measures the extent of consensus under the noise. We adjust the network parameters to change the structural asymmetries with an aim of studying their effects on the coherence. Using the graph’s structures and matrix theories, we obtain closed-form solutions of the network coherence regarding network parameters and network size. We further show that the coherence of the asymmetric graph is higher than the corresponding symmetric graph and also compare the consensus behaviors for the graphs with different asymmetric structures. It displays that the coherence of the unicyclic graph with one hub is better than the graph with two hubs. Finally, we investigate the effect of degree of hub nodes on the coherence and find that bigger difference of degrees leads to better coherence.

In this paper, we study noisy consensus dynamics in two families of weighted ring-trees networks and recursive trees with a controlled initial state. Based on the topological structures, we obtain exact expressions for the first- and second-order network coherence as a function of the involved parameters and provide the scalings of network coherence regarding network size. We then show that the weights dominate the consensus behaviors and the scalings. Finally, we make a comparison of the network coherence between the ring-trees networks and the recursive trees with the same number of nodes and show that the consensus of ring-trees networks is better than the trees since the initial state in the ring-trees networks is a ring.